Garrett D'Amore <garrett_damore@tadpole.com> writes:
>> Wouldn't this be better placed in sys/dev/todr.c, and use a config
>> attribute to bring it in (or let MD files.* files specify its use)?
>
> I don't think so.  Its not a todr itself, but rather the framework by
> which a todr interacts with the rest of the kernel.  Normally this code
> is found in some machdep kernel code next to where the clock setup is
> done, and not part of the dev/ tree for the port.

Arguably, we should use a clock device with a standard API internally
and have the inittodr/resettodr code communicate with that. The
reason inittodr/resettodr are not themselves suitable as that API is
because they do too much work that is common across ports.

So, I suppose I'm defending Garrett's placement of the code, in that
it isn't really device driver code, it is stuff that reaches to all
sorts of places including the root file system's last superblock
update time etc.

Perry